Quantifiable correlation of ToF‐SIMS and XPS data from polymer surfaces with controlled amino acid and peptide content

Peptide‐coated surfaces are widely employed in biomaterial design, but quantifiable correlation between surface composition and biological response is challenging due to, for example, instrumental limitations, a lack of suitable model surfaces or limitations in quantitatively correlating data from different surface analytical techniques. Here, we first establish a reference material that allows control over amino acid content. Reversible addition‐fragmentation chain‐transfer (RAFT) polymerisation is used to prepare a copolymer containing alkyne and furan units with well‐defined chain length and composition. Huisgen Cu(I)‐catalysed azide‐alkyne cycloaddition reaction is used to attach the model azido‐polyethyleneglycol‐amide‐modified pentafluoro‐l‐phenylalanine to the polymer. Different compositional ratios of the polymer provide a surface with varying amino acid content that is analysed by X‐ray photoelectron spectroscopy (XPS) and time‐of‐flight secondary ion mass spectrometry (ToF‐SIMS). Nitrogen‐related signals are compared with fluorine signals from both techniques. Fluorine and nitrogen signals from both techniques are found to be related to the copolymer compositions, but the homopolymer data deviate from this trend. The approach is then translated to a heparin‐binding peptide that supports cell adhesion. Human embryonic stem cells cultured on copolymer surfaces presenting different amounts of heparin‐binding peptide show strong cell growth while maintaining pluripotency after 72 h of culture. The early cell adhesion at 24 h can be correlated to the logarithm of the normalised CH4N+ ion intensity from ToF‐SIMS data, which is established as a suitable and generalisable marker ion for amino acids and peptides. This work contributes to the ability to use ToF‐SIMS in a more quantitative manner for the analysis of amino acid and peptide surfaces.

Peptide-coated surfaces are widely employed in biomaterial design, but quantifiable correlation between surface composition and biological response is challenging due to, for example, instrumental limitations, a lack of suitable model surfaces or limitations in quantitatively correlating data from different surface analytical techniques.
Here, we first establish a reference material that allows control over amino acid content. Reversible addition-fragmentation chain-transfer (RAFT) polymerisation is used to prepare a copolymer containing alkyne and furan units with well-defined chain length and composition. Huisgen Cu(I)-catalysed azide-alkyne cycloaddition reaction is used to attach the model azido-polyethyleneglycol-amide-modified pentafluoro-Lphenylalanine to the polymer. Different compositional ratios of the polymer provide a surface with varying amino acid content that is analysed by X-ray photoelectron spectroscopy (XPS) and time-of-flight secondary ion mass spectrometry (ToF-SIMS).
Nitrogen-related signals are compared with fluorine signals from both techniques.
Fluorine and nitrogen signals from both techniques are found to be related to the copolymer compositions, but the homopolymer data deviate from this trend. The approach is then translated to a heparin-binding peptide that supports cell adhesion.
Human embryonic stem cells cultured on copolymer surfaces presenting different amounts of heparin-binding peptide show strong cell growth while maintaining pluripotency after 72 h of culture. The early cell adhesion at 24 h can be correlated to the logarithm of the normalised CH 4 N + ion intensity from ToF-SIMS data, which is established as a suitable and generalisable marker ion for amino acids and peptides.
This work contributes to the ability to use ToF-SIMS in a more quantitative manner for the analysis of amino acid and peptide surfaces.

K E Y W O R D S
cell-material interaction, heparin-binding peptide, human embryonic stem cells, peptidepolymer conjugates, surface analysis 1 | INTRODUCTION Controlling cell response by modification of the physicochemical properties of the cellular environment is an attractive prospect in stem cell research as it provides the means to control stem cell fate (e.g., proliferation and differentiation). A cells' phenotypic response to its environment may initiate the expression of specific genes 1 or signalling pathways 2,3 that influence cell response. For example, osteogenesis and adipogenesis may be promoted in human embryonic stem cell (hESC) cultures by phosphate and t-butyl containing surfaces. 4 Understanding and controlling such cell-material interactions remains an active research area that requires improvement both in terms of biomaterial design and surface characterisation.
Bioactive materials harvested from native extracellular matrix such as laminin, fibrin and collagen are frequently used for cell-matrix studies as they are highly cell adhesive 5 ; however, these types of naturally derived biomimetic materials (e.g., Matrigel 6 ) suffer from nonnegligible batch-to-batch variations in chemical composition and mechanical properties. Peptide sequences (e.g., RGD 7 , IKVAV 8 and heparin-binding peptide [HBP] 9 ) are frequently used in synthetic biomimetic materials to mimic a specific function in a controlled manner such as cell attachment to surfaces 10 or self-renewal of cells. 11 HBPs possess the ability to bind heparin and other sulphated glycosaminoglycans within the extracellular matrix. They have been used on surfaces to bind and release cytokines and growth factors for wound healing, 9,12 and they have been shown to support long-term hESC expansion. 13 A systematic study that relates cell adhesion to quantitative surface analysis data that parametrise the amount of HBP on the surface has not yet been reported.
Quantitative surface analysis of peptide-coated surfaces is challenging, in part due to the compositional variability often inherent in biomaterial surfaces, and in parts due to the lack of well-defined model systems to provide a reference for analysis. Often, full surface coverage with the amino acid or peptide is not required or achieved 14,15 as submonolayer coverage down to 2% can be sufficient to modulate cell response. 16 These low amounts impose additional challenges on surface analysis methods as they require techniques with sufficiently high sensitivity and low detection limits. It is not uncommon in literature that successful surface modification is inferred mainly based on an observed effect on cell response to the material rather than a direct measurement.
A combination of surface analytical techniques is typically employed for the analysis of peptides on surfaces to complement the strengths and weaknesses of individual techniques. The quantitative nature and low surface penetration of X-ray photoelectron spectroscopy (XPS) and-in some instances-high sensitivity of time-of-flight secondary ion mass spectrometry (ToF-SIMS) make these two techniques a frequent choice for combined biomaterial surface analysis. 15,17 Other techniques such as infrared 18 and surface-enhanced Raman spectroscopy 19 have also been explored, either alone or in combination with XPS or ToF-SIMS.
XPS is able to provide quantitative information on the relative amount of amino acids or, in some cases, peptides present on the surface. The elements or chemical functionalities of interest must, however, be clearly resolved from the underlying material surface, not overlap with each other and be present in sufficient amounts to be detected (typical XPS detection limits are 1 to 0.1 at%, but can be better than 0.01 at% for elements with high photoionisation cross sections in a light-element matrix, e.g., fluorine in organic materials 20 ).
Using powder samples of the amino acids Gly, Asp, Glu, His and Arg as references and the peptide sequences RGD, RGDS and RGDSC as model samples, Stevens et al showed that both qualitative and quantitative measurement of the amount and composition of peptide sequences is possible by XPS if the amino acids are sufficiently distinct from each other. 21 Although this represents a significant advance in the data that can be extracted from XPS spectra of peptide surfaces, the approach is not universal for all amino acids. It requires correction for organic contaminants present on the surface and relies on sufficient signal strength from the amino acids and the ability to resolve and distinguish signal contributions from the substrate and the peptide that will be more difficult to achieve on small amounts of surface bound peptides than on powder samples.
ToF-SIMS has proven to be well suited to detect low amounts of amino acids or peptides on surfaces. Low surface amounts (submonolayer coverage) of single amino acids (e.g., Phe, Gly and Leu 22 ) and peptides such as RGD, 23,24 RGDS 25 and phosphorylated RGDS 25 were detectable by ToF-SIMS and with the recent development of MS-MS capabilities for SIMS, proteins can also be identified. 26 Increasingly sophisticated tools such as machine learning algorithms 27 emerge to aid qualitative SIMS data interpretation and identification of amino acids and peptides. Due to the matrix effect, that is, the dependence of secondary ion intensities on the presence of other materials in the sample ion intensities derived from ToF-SIMS are generally not considered quantitative. 28 In some circumstances, it is possible to use ToF-SIMS data quantitatively with the help of a correction parameter that can be determined from samples for which the matrix effect is well understood to and accounts for the magnitude and sign of the matrix effect on a specific ion. 29 Attempts were made in the past to connect XPS data with ToF-SIMS data to enable more reliable quantification via ToF-SIMS, for example, to determine the relationship between thiol solution and surface layer composition in the formation of mixed self-assembled monolayers 30 or to interrogate the internal chemical distribution of submicron polymer-based particles. 31 In a study that compares selected ion intensity ratios from ToF-SIMS with the O/C ratio or the %COOH obtained from oxidise polyethylene, no conclusive correlations between XPS and ToF-SIMS data were found. 32 In contrast, calibration of principal components analysis (PCA) results from ToF-SIMS with the C/O ratio obtained by XPS from a plasma-treated polypropylene samples via partial least square regression showed some promising correlations. 28 Beer et al investigated the performance of XPS, ToF-SIMS and enzyme-linked immunosorbent assay (ELISA) for the quantification of amino acid and peptide ligands on a polymer hydrogel film and found that ToF-SIMS demonstrates higher sensitivity than XPS, allowing the use of ToF-SIMS to detect peptides at biologically relevant, low quantities that are not accessible by XPS. 33 The polymer system used was an end-chain functionalised three-arm polyether that was formed into a hydrogel film, in which the amount of peptide ligand was modulated by varying the relative solution composition of polymer versus ligand.
In this study, we firstly address the challenge of developing a suitable amino acid-containing reference material to study XPS and ToF-SIMS data correlation. As many polymer-based biomaterials employ side-chain functionalisation in copolymers, we decided to develop a random copolymer-based system where ligand functionalisation is not solution controlled but modulated through the copolymer composition. Using controlled reversible addition-fragmentation chain-transfer (RAFT) copolymerisation and copper-catalysed azide-alkyne cycloaddition (CuAAC) click chemistry, we prepared a model surface that contains the amino acid Fmoc-pentafluoro-L-phenylalanine (pf-F). This amino acid contains fluorine atoms that are readily identifiable in both XPS and ToF-SIMS spectra and serve as internal reference of the model amino acid surface. Secondly, we used this model surface to investigate if a correlation can be established between the XPS and ToF-SIMS data from this model sample. Finally, the approach was translated to a biologically relevant system, an HBP-containing copolymer, to test the existence of quantifiable correlations between surface analysis data (XPS and ToF-SIMS) and cell expansion.   Twenty-five milligrammes were obtained as a pale yellow gum.
GPC was not taken due to poor solubility. The compound was characterised by NMR ( Figure S18), and the films were analysed with XPS and ToF-SIMS.

| X-ray photoelectron spectroscopy
Surfaces were analysed using a Kratos (Manchester, UK) Axis Ultra XPS instrument featuring a monochromated Al K α X-ray source pro-  Table S2); fluorinated carbons (C F; no constraints) and carboxylic acids/esters and amides (C(═O)OX/ C(═O)N where X is either H or C; no constraints). All peaks were fitted using a GL (30) function.
Peak fits carried out on the homopolymers (P3 (pf-F 0%) and P3 (pf-F 100%)) showed that the binding energy shift of aromatic carbons and amines/esters depended on the polymer composition. Although the energy separation between the C C/C H and C═C components is small such that these components would normally be fitted using a single peak, here we found that use of a single C C/C H/C═C peak reduces reproducibility of the fitting procedure. Fitting C C/C H and C═C in separate peaks increased reproducibility and allowed application of the same fitting parameters and constraints to all polymer surfaces. For the copolymers, the peak position for the C═C and C N/C O C components was therefore fixed to binding shifts that were determined by weighting the binding shift contribution using the binding energy shifts of the homopolymers and the theoretical composition of the copolymers. The resulting shifts are reported in Table S2, and the results of the fitting were tested by plotting the per-

| Embryonic stem cell attachment and expansion
Substrates in 24-well plate format were washed three times with 70% ethanol followed by three sterile PBS washes. Following air drying, substrates were exposed to mouse embryonic fibroblast

| Statistical analysis
Trendlines for all data were fitted and the coefficients of correlation, Such a comparison and correlation of data from XPS and ToF-SIMS measurements requires samples that display a large degree of uniformity in their surface composition (i.e., amino acid or peptide density) both laterally and within the analysis depth probed by both instruments. We therefore designed a copolymer system that could translating the procedures to a more biologically relevant ligand, a HBP with the sequence GKKQRFRHRNRKG ( Figure S1).
Copolymer P1 was synthesised from a trimethylsilyl-protected alkyne containing methacrylate (1) and a furan-modified methacrylate using RAFT polymerisation (Figure 1). 37 The second monomer has previously been shown to be biocompatible and acts as a filler to control copolymer composition, whereas the alkyne side group in the first monomer provides convenient anchor points for further modification with amino acids or peptides via CuAAC click chemistry after removal of the protecting group (P2 and P3). The polymerisation kinetics were determined ( Figure S2) and used to calculate the required reaction time for 80% conversion. Synthesis of the polymers was confirmed by NMR ( Figure 2) and GPC ( Figure S3). GPC analysis of copolymer P1 F I G U R E 2 1 H NMR spectra showing the stepwise side chain modification of the n = m = 41 polymer. Peak labels (a-g) represent protons in differing chemical environments from P1 to P3 (pf-F 50%), the corresponding structures of which are shown to the right of the spectra. P1-P3 (pf-F 50%) TMS reference peak at 0 ppm is represented by x, as is CH 2 Cl 2 at 5.32, and CHCl 3 is shown at 7.26 in P1 and P2. The three spectra were recorded in CDCl 3 As surface analytical techniques and NMR provide different information and have different sensitivity and detection limits, we carried out XPS analysis on films of P1, P2 and P3 (pf-F 50%) to investigate if any residual signals related to incomplete conversion could be detected ( Figure S6 and Table S1). Deprotection of P1 to P2 was evident by the drastic decrease of the Si2p signal originating from the trimethylsilyl group, but traces of Si were still present on the P2 and P3 films. Although this suggests that either deprotection was not complete or that some signals from the underlying glass substrate may be present in the spectra, the total amount of Si in the final material is very low and can be assumed to have little effect on the surface analysis. The presence of pf-F in P3 (pf-F 50%) was indicated by the appearance of F1s and N1s signals, which are uniquely introduced in the sample by the model amino acid derivative.
As copper salt(s) may be introduced by the click reaction, and its presence on the films may affect biological responses, removal of copper is essential. No significant amounts of copper were detected by XPS on any samples, indicating that purification of the polymers was successful in removing copper ions.

| Comparison of ToF-SIMS and XPS data from surfaces with varying amino acid content
XPS and ToF-SIMS were used to obtain qualitative and quantitative information from the model amino acid films to answer two questions.

| Theoretical versus experimental composition determined by XPS
XPS spectra of the polymer-amino acid conjugate films P3 (pf-F 0-100%) are shown in Figure 3. Carbon (C1s) and oxygen (O1s) signals originate from both the pf-F side group and the copolymer backbone and are present in all spectra. Increasing amounts of pf-F in the film were accompanied by increasing F1s and N1s peak intensities for a pf-F content up to 50%. For the 100% pf-F sample, both the F1s and N1s peak intensities decreased below that of the 50% pf-F sample (Table S1). The amounts of silicon detected were generally very low and comparable with the levels detected on the control sample (P2), suggesting that no significant contributions to the spectra can be expected from the glass substrate.
To further confirm the chemical composition of the films, peak fitting was carried out on the high-resolution C1s spectra ( Figures S6   and 3). Due to the complex and varying composition of the spectra related to the diverse chemical environments present in the polymeramino acid conjugates, a number of chemical environments (amines and ethers; carboxylic acids, esters, amides and carbamates) could not readily be distinguished from each other and were combined into single fitted peaks. In addition, the fitted components clearly showed the presence of C F (from pentafluoro-L-phenylalanine) and C═C (from Fmoc, pentafluoro-L-phenylalanine and furan). As the binding energy shifts for similar chemical environments are different for the two monomer units, the peak positions were affected by the copolymer position, requiring imposition of some constraints that take the varying composition into account. The constraints used are provided in Table S2 and were derived based on the chemical shifts observed in the homopolymers, which were adjusted proportionally based on the theoretical composition of each copolymer. The resulting relative amount of each fitted component is provided in Table S3. The goodness of fit was tested by plotting the amount of C F obtained from the component fitting against the relative amount of F1s detected in the wide scan spectra ( Figure S7). The correlation fits a linear trendline (R 2 = 0.984) and has a statistically significant (ANOVA, p = 0.057) positive association (Pearson's r value = 0.996, Table S15) between the amount of F from the survey spectra and the amount of C F from the fitted high-resolution spectra, suggesting that the fitting parameters are appropriate.
The types of components fitted match the expected chemical environments very well. A continuous increase in the amount of C F and C═C bonds was observed when the pf-F content increased from 0 to 100%, except for the P3 (pf-F 100%) where the amount of C F dropped compared with the P3 (pf-F 50%) sample (Table S3). However, when compared with the theoretical values, it is apparent that the relative amount of C F is lower than expected, which matches the observation of an overall decrease in F1s peak intensity in the survey spectra.
To determine how the experimental XPS data matches expected theoretical compositions and understand the unexpected behaviour of the P3 (pf-F 100%) sample better, N/C, F/C and F/N atomic ratios were determined both from experimental XPS data and from the theoretical (co)polymer compositions (Table S4). The F/N ratio should be close to 1 for all pf-F containing samples. The average experimental values deviate slightly from the expected ones, but this can in general be attributed to measurement errors, in particular for samples where very low amounts of pf-F are present.
The N/C and F/C ratios can be used as a measure of how well the experimental data fits the expected polymer composition ( Figure 4). Comparison of the experimental XPS data with the theoretical pf-F content ( Figure 4A) suggest an increasing trend for the three copolymer samples (P3 (pf-F 1%), P3 (pf-F 10%) and P3 (pf-F 50%)) but the homopolymers (P3 (pf-F 0%) and P3 (pf-F 100%)) deviate from linearity. A similar trend was observed when plotting experimental and theoretical values for the N/C and F/C rations and the relative amounts of C F and C═C bonds ( Figure S8).
The percent of pf-F relates to the number of repeat units in the polymer and is therefore not a direct representation of the amount of material (pf-F vs. furan) present in sample. This is of particular importance later on when the mass of the two repeat units (HBP peptide vs. furan) is even larger than for pf-F and furan (Table S5).
We therefore converted the theoretical and experimental polymer for the P3 (100% pf-F) sample suggest that considerably less pf-F was present in the analysed sample volume than expected. The conversion of the alkyne side groups to azides and hence attachment of pf-F was confirmed by 1 H NMR. However, even though the conversion appeared complete by NMR, the sensitivity of this technique is considerably smaller than that of XPS, and XPS is, therefore, more likely to be affected by incomplete functionalisation. This is particularly true for the homopolymer, as it has been shown before that steric hindrance can prevent full functionalisation of polymer side chains. 38

| Linearity and correlation of ToF-SIMS secondary ion intensities with amino acid content
Reliable Images generated from these ions showed uniform chemical distribution ( Figures S13 and S14). In this data, it is noticeable that the P3 (pf-F 0%) sample contains considerable amounts of sodium, whereas the intensity of the other ions is significantly decreased compared with the copolymer samples. It is possible that this is caused by the difference in the synthetic procedure, as trimethylsilyl deprotection and CuAAC click chemistry were not carried out on this sample.
The normalised intensity of all identified marker ions is plotted against the pf-F content of the sample in Figure S15; a representative selection is shown in Figure 5A. The most robust secondary ion for the monitoring of different pf-F concentration in the polymer films with ToF-SIMS is F À , as its intensity is relatively strong and increases continuously with pf-F content, and the standard deviations for the data points are small.
Although F À is a unique marker ion for the model amino acid pf-F, it is not a suitable universal marker ion for peptides. In contrast, nitrogen-containing fragments generated in ToF-SIMS spectra (CN À , CNO À and CH 4 N + ) are rather generic but can be detected in all peptides. To investigate how suitable these three nitrogencontaining ions are to follow different amounts of peptide on a surface by ToF-SIMS, the F/N ratio (determined by XPS) and the ratio of the F À ion over each of the three nitrogen-related ions (determined by ToF-SIMS) were plotted against the pf-F content of the samples ( Figure S16). These ratios are expected to be independent of the polymer composition, as they only use signals present in pf-F.
They can therefore be used as an indicator of how well these nitrogen-containing ions relate to the amount of amino acids/ peptides present on the surface. The F/N ratio (XPS) was already F I G U R E 4 Correlation between experimental XPS data and theoretical composition of P3 (pf-F) films. (A) F/C and N/C ratios (determined by XPS) as a function of theoretical pf-F content in the polymer films. (B) Mass fraction of pf-F determined from either the F/C or the N/C ratios obtained by XPS and their correlation with the theoretical mass fractions of pf-F. Lines represent linear fits for the data points of the copolymer samples only. Homopolymer samples (0 and 100% pf-F) were excluded from the line fit as they showed behaviour that significantly deviated from that of the copolymers, likely due to the difference in the material synthesis and composition. Errors (included but very small in value) are standard deviations from repeat measurements (n = 6). XPS, X-ray photoelectron spectroscopy discussed above and shows matching values for polymers with a pf-F content equal or higher than 1% when allowing for the experimental error of the P3 (pf-F 1%) sample ( Figure S16A). The F À /CN À and F À /CNO À ratios increase when the pf-F content increases up to 50%, but then drops for the P3 (pf-F 100%) sample ( Figure S16B).
Assuming the F À ion intensity to be a robust indicator of pf-F content, this indicates that the CN À and CNO À ion intensities are either not solely dependent on pf-F concentration or that they display a different matrix effect character compared with the fluoride ion. In contrast, the F À /CH 4 N + ion ratio is, within experimental error, constant for all pf-F-containing polymers. CH 4 N + was therefore selected in addition to F À for further analysis as a more general marker ion for peptides. As a marker for the furan unit, the C 4 H 3 O À ion was chosen over the C 2 H 5 O + ion for subsequent data analysis because it displayed less variability (highest standard deviations are 9% and 22% for C 4 H 3 O À and C 2 H 5 O + , respectively).
Normalised secondary ion intensities for pf-F (F À , CH 4 N + ) and furan (C 4 H 3 O À ) associated ions are plotted in Figure 5A. The normalised intensities of F À (amino acid) increases constantly while that of CH 4 N + (amino acid) drops from P3 (pf-F 0%) to P3 (pf-F 1%) but then increase continuously with increasing pf-F content.
Normalised C 4 H 3 O À ion intensities also follow the expected trend of continuous decrease with increasing pf-F content, with the exception of the P3 (pf-F 0%) sample, that has considerably lower C 4 H 3 O À ion intensities than the P3 (pf-F 1%) sample.
To relate ToF-SIMS secondary ion intensities with the amount of amino acid on the surface, the ratio of the normalised marker ion intensities was plotted in Figure 5B against the mass fraction of pf-F determined by either using the N/C or the F/C ratio measured by XPS (Section 1. 2.3 and Tables S7 and S8). The homopolymer samples display deviating behaviour from the copolymer samples in line with the observations made for the XPS data. Linear regressions for the copolymer datasets were generated both for nonlogarithmic and logarithmic plots; using the coefficient of determination (R 2 value, Table S15), it was found that the F À and CH 4 N + ions (amino acid markers) produce the strongest correlation with the experimental mass fraction of pf-F in a log-log plot ( Figure 5B) while the C 4 H 3 O À and C 5 H 5 O + ions (furan markers) produce a better fit to a power law if the normalised secondary ion intensity is plotted on a log scale ( Figure S17).
These linear relationships between the logarithm of normalised secondary ion intensities from ToF-SIMS and the logarithm of the mass fraction of pf-F determined from XPS data is strongest for the F À versus F/C data pair and weakest for the CH 4 N + versus N/C data pair (Table S15). The stronger association when fluorine-related data (plots involving F À or F/C values) are used is likely due to the decreased likelihood of contamination or interference from other species with fluorine-related signals. Even though the correlation between nitrogen-based XPS and ToF-SIMS (CH 4 N + and N/C related) data is statistically slightly weaker than that of fluorine based datasets, the parameters of the regression line, particularly the slope (b), are very similar, suggesting that signals from either fluorine or nitrogen components can be used for the XPS/ToF-SIMS correlation with similar accuracy. Hence, even if the nitrogen related data displays reduced precision, the data in Figure 5 and Table S15 supports its use as a marker for amino acids and peptide surfaces that can be more generically applied than the more specific but artificially introduced fluorine marker. tions. 13 We hypothesised that polymer films with varying HBP content would result in differences in the adhesion of hESCs that can be quantitatively described by correlation of XPS and ToF-SIMS data of the surface acquired before cell culture to explore the ability to use ToF-SIMS data for quantitative correlation in biology.

| Preparation of HBP-containing polymers
To prepare HBP-containing polymers, the same alkyne and furan containing copoylmers designed for the previous pf-F study (P2) were used and functionalised with an azide containing HBP ( Figure S1).
As the response of hESCs has previously been shown to be enhanced by small quantities of surface bound peptides, 13 polymers in which HBP makes up 1%, 5% or 10% of the side chains were prepared. Homopolymers were not included in this study as the pf-F study showed that the homopolymers deviate significantly from the linear correlations observed for the copolymers.
The HBP-containing copoylmers were analysed by NMR ( Figure S18). The 1 H NMR showed an increase of peptide-binding domain signals from 1% to 10% HBP and a decrease of the furan associated peak. Due to the complexity of the HBP spectrum and the overlay of HBP signals with most of the copolymer signals, no further analysis of the NMR data was undertaken.

| XPS analysis of HBP-containing polymer films
XPS spectra of the P3 (HBP) samples showed the presence of carbon, oxygen and nitrogen, but also strong signals for fluorine ( Figure S19 and Table S10). As neither the HBP peptide nor the polymer contains fluorine, the most likely reason for the presence of fluorine is residual trifluoroacetic acid, which was used in the synthesis of HBP to remove the side chain protection groups and subsequently be present as the counter anion to the basic amino acid residues in HBP.
Following the increase in the number of nitrogen present with increasing amount of HBP in the polymer, the relative amount of nitrogen measured by XPS increased from P3 (HBP 1%) to P3 (HBP 10%) (Table S10). This indicates that the HBP modification was successful and that the samples indeed display increasing amounts of HBP.
The theoretical and experimental N/C ratio was computed (Table S11) and plotted against each other in Figure S20. The experimental values show very high correlation (Pearson's r value = 0.999) with high statistical significance (ANOVA, p value = 0.009) (Table S15) Table S12, results of the calculations are given in Table S13 and the final plot is shown in Figure S25. Although a linear trendline with acceptable fit (R 2 > 0.85) and high correlation (Pearson's r value > 0.96) can be placed through the data, the correlation is not statistically significant (ANOVA, p > 0.1) ( hESCs were cultured on the samples and cell counts were performed after 24 and 72 h ( Figure 6 and Table S14). Cell numbers were statistically different between all samples at both timepoints, except for the P3 (HBP 1%) and P3 (HBP 5%) samples, for which the difference was not statistically significant.
The hESCs densely populate the Matrigel surface as well as the P3 (HBP 5%) and P3 (HBP 10%) samples ( Figure 6A). On the P3 (HBP  Figure S26) shows that hESCs maintain pluripotency, supporting the assumption that the observed increase in cell count from 24 to 72 h is due to cell expansion rather than cell differentiation.  When investigating linearity and association between surface analytical data, the homopolymer samples consistently displayed deviations from linearity compared with the copolymer samples. It must therefore be concluded that although homopolymers may still be suitable reference materials for biological experiments, when aiming to establish systematically varying trends of physicochemical properties, the polymer system we designed only operates well if used as a copolymer. This finding could be of relevance to other polymer systems, and we advise that care should be taken when combining homopolymer and copolymer surface analysis data for the investigation of systematic trends.
Our copolymer system provided a suitable platform to study the influence of varying amounts of HBP on the maintenance of selfrenewal capability of hESCs. hESCs maintained their self-renewal capability on the HBP-containing copolymers over 72 h of culture.
In addition, cell adhesion increased with increasing HBP content in the copolymer, which was shown to correlate to the logarithm of the normalised CH 4 N + ion intensity measured by ToF-SIMS with high statistical significance. The system developed here may provide opportunities towards a scalable platform for the expansion of hESCs to improve hESC availability for subsequent biomedical applications.